Who`s on Top? The Mental Health of Men Who Have Sex with Men

INQUIRY
Volume 18
Who’s On Top? The Mental Health of Men Who Have Sex with Men
By: Eric R. A. Carter
Department of Psychological Science
Faculty Mentor: Dr. Ana J. Bridges
Department of Psychological Science
Abstract
Despite most men who have sex with men (MSM) expressing intercourse position preference (e.g., “top”,
“versatile”, or “bottom”), there is little information regarding sexual behavior and mental health sequelae.
From the perspective of gender schema theory, the current study examined how position preference related to
gender roles, internalized homophobia, and mental health. A total of 70 MSM (U.S. residents, M age = 28.89
years, 68.6% White) were recruited for an online study and grouped according to position preference. Groups
were mostly similar across demographic variables, although bottoms had fewer sexual partners and lower
condom use than tops and versatiles. In terms of gender roles, tops and versatiles were significantly higher
in both masculine and feminine traits than bottoms. Tops were significantly more likely to report internalized
homophobia than versatiles and bottoms. After controlling for masculinity, versatiles had the highest mental
health. Results suggest further study of different sub-populations of MSM is warranted.
Keywords: anal sex, gender roles, homophobia, mental health
As tolerance toward homosexuality has
increased in the U.S., so too has research regarding
the health and mental health of homosexual
populations. One of the most researched areas
regarding the health of homosexual populations,
particularly in men who have sex with men (MSM),
is sexual disease risk. There is a wealth of research
on the habits, behaviors, and mental health of MSM
afflicted with human immunodeficiency virus (HIV)
or acquired immune deficiency syndrome (AIDS)
(Baggaley, White, & Boily, 2010; Koblin et al.,
2006; Mustanski, Newcomb, Du Bois, Garcia, &
Grov, 2011; Safren, Reisner, Herrick, Mimiaga,
& Stall, 2010). Much of the research has focused
on the sexual behaviors (e.g., penetrative anal sex
without the use of a condom) that increase the risk of
transmission of sexually transmitted infections (STI),
especially HIV (Jin et al., 2009; Koblin et al., 2006;
Thomas et al., 2009; Van Druten, Van Griensven, &
Hendriks, 1992).
Two behaviors that have been examined in
relation to HIV and sexually transmitted infection
(STI) risk consist of engaging in either receptive or in
insertive anal intercourse (Jin et al., 2009; Thomas et
al., 2009; Van Druten et al., 1992). Those who engage
primarily in insertive anal intercourse are referred to
as adopting an active, top, or insertive role; those who
engage in receptive anal intercourse are often referred
to as adopting a passive, bottom, or receptive role.
Some men engage in just one anal sex role (receptive
or insertive), while others are more versatile in their
behavior (Patterson & D’Augelli, 2013). Nearly 90%
of MSM identify with a self-label of “top,” “bottom,”
or “versatile” (Hart et al., 2003; Wei & Raymond,
2011). Prevalence rates are approximately 35% top,
25% bottom, and 40% versatile (Grov, Parsons, &
Bimbi, 2010; Wei & Raymond, 2011). HIV and other
STI risks are increased by adhering to a bottom anal sex
role during intercourse (Wegesin & Meyer Bahlburg,
2000). A meta-analysis found that per-partner risk of
HIV infection during unprotected sex was 40.4% for
bottoms while risk was only 21.7% for tops (Baggaley
et al., 2010).
In addition to sexual disease risk, some
researchers have examined whether anal sexual
preferences are associated with specific gender roles.
For example, preferring a bottom role is often associated
with being passive or feminine, while preferring a top
4
PSYCHOLOGICAL SCIENCE: Eric Carter
role is associated with being aggressive and masculine
(Wei & Raymond, 2011). Weinrich, Grant, Jacobson,
Robinson, and McCutchan (1992) found that
preference for bottom roles in 102 MSM residing in
a large west coast city was associated with childhood
feminine gender expression. Kippax and Smith (2001)
examined power relationships and conceptualization
of MSM’s intercourse position preference and
relationships, finding bottom preference is associated
with being passive, weak, and feminine, while top
preference is associated with being active, strong, and
masculine.
In many relationships among MSM,
masculinity is often viewed as ideal, whereas femininity
is considered undesirable (Lanzieri & Hildebrandt,
2011; Patterson & D’Augelli, 2013). This mirrors
the general societal preference for masculinity over
femininity in Western cultures (Connell, 2005). For
example, Bailey, Kim, Hills, and Linsenmeier (1997)
investigated MSM’s romantic partner preferences
by examining characteristics of ideal partners that
men listed on a prominent dating website. In this
investigation, masculinity was the most commonly
listed desirable trait in a potential partner. Furthermore,
no single profile listed masculinity as an undesirable
trait. Feminine traits, however, were commonly listed
as undesirable. Because masculinity is a valued trait
in men and femininity often is not, and because
bottom preference is associated with femininity, it is
possible that men who prefer bottom roles and who
are considered not very masculine may be at risk for
depression or other mental health problems. Indeed,
Rieger and Savin-Williams (2012) demonstrated
gender nonconformity (i.e., high femininity and low
masculinity) was negatively related to psychological
well-being in 475 high school boys.
According to gender schema theory, people
learn about how to be “men” and “women” from a
complex socialization process that includes parenting,
schooling, mass media, and peer group interactions
(Bem, 1981). However, individuals differ in how
much they conform to society’s typical “masculine”
or “feminine” gender roles. People can be categorized
according to gender conformity types. For example,
“masculine” men and “feminine” women are defined
as sex-typed. “Feminine” men and “masculine”
women are defined as cross-sex-typed. Men and
women who are high on both feminine and masculine
traits are considered androgynous. Finally, men and
women who are low on both feminine and masculine
traits are considered undifferentiated.
Many studies have pointed to the benefits of
a masculine gender role orientation and the relatively
detrimental effects of a feminine orientation for the
mental health of both men and women (e.g., Bassoff
& Glass, 1982; Cella, Iannaccone, & Cotrufo,
2013; Taylor & Hall, 1982). Alternatively, there is a
sizeable literature suggesting people who are more
androgynous in their gender roles, endorsing both
masculine and feminine traits, have the best mental
health of all (e.g., Cheng, 2005; Coleman, Kaplan,
& Casey, 2011; Lefkowitz & Zeldow, 2006; Prakash
et al., 2010; Woodhill & Samuels, 2003). In part
this may be because such individuals have greater
psychological flexibility and are able to be either
more assertive or more cooperative, depending on
the demands of specific situations. It may also be
that in patriarchal societies that value masculinity
over femininity, people (but especially men) who
fail to embody valued masculine traits are subject
to more discrimination than people who display
high masculinity. Indeed, Gordon and Meyer (2007)
found that gender noncomformity, above and beyond
sexual orientation, was associated with experiences
of discrimination in gay, lesbian, and bisexual
participants. In addition, Sandfort, Melendez, and Diaz
(2007) found experiences of discrimination mediated
the association between gender nonconformity and
poor mental health in a large sample of Latino gay and
bisexual men.
In addition to the potentially negative mental
health effects of gender nonconformity (in the form
of high femininity and low masculinity), some MSM
may also have high levels of internalized homophobia
(Williamson, 2000). Studies have found that gay
men with higher rates of internalized homophobia
report lower relationship satisfaction and duration,
decreased openness about sexual orientation (Ross &
Rosser, 1996), increased psychological problems such
as substance abuse, increased self-injurious behaviors
(Williamson, 2000), and higher depression (Herek,
Cogan, Gillis, & Glunt, 1998) than gay men with little
or no internalized homophobia.
In sum, the research to date demonstrates a
bottom anal sex role preference in MSM is associated
with increased sexual health risk and increased gender
5
PSYCHOLOGICAL SCIENCE: Eric Carter
nonconformity compared to top sex role preference.
Also, studies find both gender noncomformity and
internalized homophobia are negatively associated
with mental health in MSM. However, much of the
research with MSM has focused on HIV risk or HIVpositive samples, limiting generalizability. In addition,
studies have not focused much attention on top or
versatile intercourse position preferences in MSM as
they relate to mental health, perhaps because so much
of the work has been focused on sexual health risk and
the risk of HIV transmission is lower in this group.
Finally, the constructs of anal sex role preference,
gender role orientation, internalized homophobia, and
their mental health sequelae have not been investigated
together in a single sample of men.
Purpose­­
The first exploratory aim of this research was
to determine demographic characteristics associated
with anal intercourse position preference. The second
aim was to examine how position preference related
to internalized homophobia and gender roles, and how
these in turn related to mental health. Consistent with
gender role schema theory, we anticipated that gender
congruence (i.e., masculinity) would be highest in
men who identified as “top” or “versatile” (Hypothesis
1), while non-congruence (i.e., femininity) would be
higher in men who identified as “bottom” (Hypothesis
2). We further anticipated both versatile and bottom
position preferences would be associated with lower
internalized homophobia, compared to a top preference
(Hypothesis 3). Finally, we expected a gender
non-congruent orientation (i.e., higher femininity;
Hypothesis 4) and higher internalized homophobia
(Hypothesis 5) would be associated with lower global
mental health, while gender typical (sex-typed) and
androgynous gender roles would be associated with
greater global mental health.
Method
Participants
A total of 105 individuals were recruited for
participation in this study. Of these, 27 were excluded
due to lack of completion or for not meeting requisite
demographic criteria (i.e., having had a sexual
experience with another man, male gender, age > 17
years, and resident of the United States). Of the 78 who
completed the survey, eight participants consistently
selected “I do not use labels” on questions assessing
anal intercourse position preference. Given this was
the primary grouping variable used in the study, these
participants were also excluded from analyses. Thus,
the final sample size used in the statistical analyses
and comparisons consisted of 70 men.
The average age of the final sample was
29.89 years (SD = 8.43, range 20-55). In terms of
race, 68.6% of the sample self-reported being White.
A total of 47.1% of the sample reported adhering to
a religion. The average educational level was 4.69
(SD = 1.56), ranging between 4 (some college) and
5 (undergraduate degree). Fifty-seven percent of
participants were employed full-time, with an average
annual income between $25,000 and $49,999. About
57% of participants reported being in a relationship,
with an average duration of 3.32 years (SD = 4.63,
range one month – 20 years).
Procedures
A majority (86%) of participants were recruited
via MechanicalTurk, an internet-based recruitment
and participant payment service. The recruitment
message informed participants that they were going to
complete a survey concerning the sexual behavior of
MSM. As per MechanicalTurk’s policy, the recruitment
message specified that the content of the survey was
appropriate for adults only and filters were put in place
that required participants to have already verified their
age and willingness to see survey requests such as this;
individuals younger than 18 were not able to see the
recruitment message. Participants were also informed
they would receive a modest monetary compensation
for completing the survey.
In addition to being MSM, aged 18 and over,
and living in the United States, participants also had
to meet MechanicalTurk criteria as “Master Workers”.
This elite status indicated that the participants in the
study had a high degree of accuracy and completion
of work and a high approval rate from recruiters on
other MechanicalTurk tasks. After meeting all of the
criteria, the recruited participants were provided with
a short memorandum about the study and a link to
the SurveyMonkey questionnaire itself. At the end of
the study, participants were given a code that enabled
them to receive payment via MechanicalTurk. In order
to enhance recruitment, we increased the amount of
compensation from the initial $1 to $2, $3, and finally
$4. Thirty-two participants received $4, 5 received $3,
3 received $2, and 27 received $1.
Fourteen percent of participants were recruited
via snowball sampling. In particular, the researcher
and his advisor contacted acquaintances who met
6
PSYCHOLOGICAL SCIENCE: Eric Carter
criteria and asked if they would participate and
recruit others who met criteria to also take the survey.
Because they were not affiliated with MechanicalTurk,
none of the participants recruited in this method were
compensated.
Once
all
participants
accessed
the
questionnaire, they were first provided with
information about the study and indicated their
consent by clicking on a button. Next, participants
were presented with a series of questions (provided
below). Once participants completed the questions,
a debriefing page was presented. Participants were
given additional information regarding the purpose of
the study and they were thanked for their participation.
All study procedures were approved by the
University of Arkansas Institutional Review Board.
Measures
Demographic characteristics. Demographic
questions were asked concerning age, ethnicity,
religion,
religiosity,
educational
attainment,
employment status, annual income, relationship
status, sexual orientation, sexual satisfaction, degree
of openness regarding sexual orientation, condom use,
and pornography use. Age was a free response and
was coded continuously. Ethnicity was a multi-select
item that was recoded into a dichotomous White/
non-White item. Multiple options (single select)
were offered for religious denomination; however
this variable was recoded into a dichotomous atheist/
theist item. Religiosity was coded from 1 (not at all
religious) to 5 (extremely religious). Educational
attainment was coded 1 (no high school), 2 (some high
school), 3 (high school diploma), 4 (some college), 5
(undergraduate degree), 6 (some higher/graduate
education), 7 (masters level degree), and 8 (doctoral
level degree). Employment status responses were
recoded into a dichotomous item: employed full-time/
not employed full-time. Annual income was coded
in $25,000 increments from 1 (less than $25,000
per year) to 7 (greater than $150,000 per year).
Relationship status was recoded into a dichotomous
variable: single/partnered. Sexual orientation was
measured by self-reported interest in one or both sexes
1 (sexual interest in women only) to 5 (sexual interest
in men only). Sexual satisfaction was coded from 1
(very dissatisfied) to 5 (very satisfied). Openness with
others regarding sexual orientation was recoded into
a dichotomous item: closeted/out. Condom use was
coded from 1 (never, 0% of the time) to 5 (always,
100% of the time). Pornography use was coded from 1
(never) to 7 (multiple times per day).
Anal intercourse position preference. Since
all previous studies used a single question to establish
anal intercourse position preference (IPP) and this
study’s primary measure of interest was IPP, we
created a six-item questionnaire that assessed IPP.
Items on the questionnaire began by asking In the past
12 months and inquired about past anal sex behavior
(When I have had sex, I have usually been…), fantasy (I
have primarily fantasized about being… and If I could
have any sort of anal sex I wanted, I would want to…),
self-labelling (I think of myself as a… and I use the
self-label of…), and others’ labelling of the participant
(People I know or have had sexual encounters with
know me as a…). The response options for the six
questions were: (1) top (inserter, pitcher, penetrator),
(2) versatile top (usually “top,” but occasionally
bottom), (3) versatile (“vers”, flip flop, doing both in
nearly equal amounts), (4) versatile bottom (usually
“bottom,” but occasionally top), and (5) bottom
(receiver, catcher, penetrated). These response choices
included multiple terms of reference (slang, etc.) for
the IPP labels so that a participant’s personal label of
choice would likely be listed within one of the five
choices. Participants were also given the options of
indicating “these labels don’t apply to me,” and “other
(please specify).”
Inter-item correlations between the six items
ranged from r = .665 to .981, and all were significant
at p < .001. Cronbach alpha for the six-item scale was
.96. Given the high internal consistency, an average
IPP score was calculated for each participant. Scores
ranged from 1 (top) to 5 (bottom). The mean score was
then recoded into three groups. Scores ranging from
1 to 1.99 were recoded as top, scores from 2.00 to
3.99 became versatile, and scores of 4.00 to 5 became
bottom.
Gender roles. The Bem Sex Role Inventory
(Bem, 1974; Hoffman & Borders, 2001) was
administered to assess participants’ gender roles.
The Bem contains 60 adjectives (1/3 of which are
typically associated with masculinity, 1/3 of which
are typically associated with femininity, and 1/3 of
which are typically not associated with one gender or
the other). Sample traits on the masculinity subscale
include self-reliant and aggressive, while sample
traits on the femininity subscale include sensitive and
compassionate. Two subscale scores, femininity and
7
PSYCHOLOGICAL SCIENCE: Eric Carter
Table 1. Demographic characteristics of the sample
by intercourse position preference group.
Top (n = 30)
M (SD) or N
(%)
Variable
Age
31.70 (7.56)
White race
18 (62%)
Versatile
(n = 28)
Bottom
(n = 12)
M (SD) or
N (%)
M (SD)
or N (%)
28.07
(8.48)
29.58
(10.08)
22 (79%)
5 (42%)
F or X2 statistic
F (2, 67) = 1.36,
p = .263
2
X (2) = 5.26,
p = .072
Atheist
13 (43%)
14 (50%)
10 (83%)
2
X (2) = 5.66,
p = .059
1
Religiosity
2.10 (1.27)
Educational
attainment2
4.77 (1.63)
Employed full time
19 (65%)
1.79
(0.92)
1.67
(1.23)
F (2, 67) = 0.86,
4.57
(1.57)
4.75
(1.42)
F (2, 67) = 0.12,
11 (41%)
10 (83%)
X (2) = 7.16,
p = .426
p = .884
2
p = .028
3
Annual income
2.37 (1.45)
4
Sexual orientation
a
2.77 (1.22)
2.46
(1.69)
2.25
(1.14)
b
c
3.50
(1.43)
4.75
(0.45)
F (2, 67) = 0.09,
p = .915
F (2, 67) =
11.38
p < .001
Single (not in a
relationship)
11 (37%)
17 (61%)
2 (17%)
2
X (2) = 7.48,
p = .024
a
Number of sexual
partners (past
year)5
3.10 (1.16)
Sexual
satisfaction6
3.77 (1.25)
7
a
Condom use
3.60 (1.43)
8
Pornography use
5.03 (1.16)
3.14
(1.27)a
2.08
(0.67)b
F (2, 67) = 5.36,
3.89
(0.97)
4.17
(1.34)
F (2, 66) = 0.51,
3.69
(1.44)
a
2.08
(1.51)
b
F (2, 65) = 5.79,
4.57
(1.10)
4.83
(1.70)
F (2, 67) = 1.00,
p = .020
p = .606
p = .005
p = .373
Note. Means sharing a letter in their superscript are not
significantly different at the .05 level on Fisher LSD posthoc tests.
masculinity, were calculated by averaging the 20 items
assessing femininity and masculinity, respectively.
Higher scores indicated greater endorsement of those
gendered traits. Studies have shown the Bem’s internal
reliability is good, with Cronbach alpha values typically
between .75 and .90 (reviewed in Hoffman & Borders,
2001). Recent evidence of content validity was
demonstrated in a study, with men being significantly
more likely to be categorized as having a male sextyped gender role and women being significantly
more likely to be categorized as having a female sextyped gender role (Marrs, Sigler, & Bramer, 2012). In
the current study, internal consistency values were .93
for the masculinity subscale and .89 for the femininity
subscale.
Internalized homophobia. The Reactions to
Homosexuality Scale, revised (Smolenski, Diamond,
Ross, & Rosser, 2010) was used to assess internalized
homophobia. This is a seven-item questionnaire
assessing three domains of homophobia: personal
comfort with a gay identity (3 items), social comfort
with gay men (2 items), and public identification as
gay (2 items). Sample items include, Even if I could
change my sexual orientation, I wouldn’t and I feel
comfortable being seen in public with an obviously
gay person. Items are answered on a 7-point Likert
scale from 1 (strongly disagree) to 7 (strongly agree).
The original scale measures positive reactions to
homosexuality, so scores were reverse-coded such
that higher scores indicated higher internalized
homophobia.
The authors demonstrated adequate internal
consistency (Cronbach alpha coefficients ranging
from .60 to .76 for the individual subscales) and
high subscale correlations (ranging from .40 to .55).
Criterion validity was demonstrated by the authors:
men who were involved in gay organizations, were
not in relationships with other men, and did not
discuss their sexual orientation with their primary
care physicians scored higher on the subscales of
internalized homophobia than men integrated in the
gay community, in gay relationships, and who openly
discussed their sexual orientation with their physicians
(Smolenski et al., 2010). In the current study, internal
consistency for the 7-item scale was .77.
Mental health. The Brief Symptom Inventory
(BSI) was administered to evaluate psychiatric distress
and mental well-being (Derogatis & Melisaratos,
1986). The BSI is a 53-item measure assessing
psychiatric symptoms across nine dimensions:
somatization, obsessive-compulsive, interpersonal
sensitivity, depression, anxiety, hostility, phobic
anxiety, paranoid ideation, and psychoticism. Items ask
about how often the respondent has been distressed or
bothered by the symptoms over the past week and are
8
PSYCHOLOGICAL SCIENCE: Eric Carter
scored on a 5-point Likert scale, from 0 (not at all) to 4
(extremely). Sample items include feeling no interest
in things and spells of terror or panic. The mean of all
items was calculated to achieve a global psychiatric
distress score. Higher numbers equal higher psychiatric
distress (worse mental health, more problems).
Internal consistency values for the symptom subscales
range from .71 to .85, with global distress total scores
showing excellent test-retest reliability over a twoweek span (rxx = .90). Convergent validity has been
demonstrated by positive correlations between BSI
subscales and theoretically similar subscales of the
Minnesota Multiphasic Personality Inventory. In the
current study, internal reliability for the global distress
score was .97.
Data from the SurveyMonkey questionnaire
were transferred into IBM SPSS Statistics program
wherein analyses of variance (ANOVA), chi square
tests for independence, and analyses of covariance
(ANCOVA) were conducted. Post-hoc tests were run
using the Fisher LSD test. For all tests, alpha levels
for evaluating statistical significance were set at 0.05.
Results
Part 1: Demographic Correlates
A series of one-way between-groups analyses
of variance (ANOVAs) and chi square tests were
conducted to explore the relation of intercourse
position preference (IPP) to numerous demographic
variables, including age, educational attainment,
annual income, religiosity, sexual orientation, sexual
satisfaction, frequency of condom use, number of
sexual partners in the past year, and pornography use.
Means and standard deviations are reported in Table 1.
The three IPP groups (tops, versatiles, and bottoms) did
not differ significantly in average age, race/ethnicity,
religious affiliation, religiosity, educational attainment,
annual income, sexual satisfaction, or pornography
use. Significant differences did emerge for full-time
employment status, with bottoms (83%) significantly
more likely to report full-time employment than tops
(65%) or versatiles (41%) (p = .028).
Regarding relationship and sex questions,
other differences between the groups emerged.
Versatiles (61%) were significantly more likely to be
single compared to tops (37%) and bottoms (17%)
(p = .024). Differences also emerged in self-reported
sexual orientation. In particular, utilizing a one-way
between-groups ANOVA, we explored the relation
between IPP and self-reported sexual orientation.
Significant differences were found between all groups
[F (2, 67) = 11.38, p < .001]. Post-hoc comparisons
using the Fisher LSD test indicated significant
differences between all three position preference
groups (all p values < .03). In particular, bottoms
reported the highest levels of homosexuality (M =
4.75, SD = 0.45). Versatiles (M = 3.50, SD = 1.43)
reported less interest in men than bottoms, but more
than tops, who reported the highest levels of sexual
interest in women (M = 2.77, SD = 1.22).
A one-way between-groups ANOVA compared
IPP groups on the number of sexual partners in the past
year. There was a statistically significant difference
among the groups [F (2, 67) = 4.13, p = .020]. Posthoc comparisons using the Fisher LSD test indicated
that the mean score for bottoms (M = 2.08, SD = 0.67,
~one sexual partner in the past year) was significantly
different from tops (M = 3.10, SD = 1.16, ~two sexual
partners in the past year) and from versatiles (M = 3.14,
SD = 1.27, ~two sexual partners) (p = .011). Tops and
versatiles did not significantly differ from each other.
There was also a statistically significant
difference in condom use between groups [F (2, 65) =
5.79, p = .005]. Bottoms (M = 2.08, SD = 1.51) used
condoms significantly less often than both tops and
versatiles. Tops (M = 3.60, SD = 1.43) and versatiles
(M = 3.69, SD = 1.44) did not significantly differ from
each other in terms of condom use.
Part 2: Hypothesis Tests
H1: Intercourse position preference (IPP)
and masculinity. A one-way between-groups ANOVA
was conducted to compare the three IPP groups on
masculinity scores, as measured by the relevant
subscale of the Bem. Table 2 provides a summary of the
results. There was a statistically significant difference
in Bem masculinity scores for the three groups [F (2,
67) = 8.57, p < .001]. Post-hoc comparisons using the
Fisher LSD test indicated that bottoms (M = 3.48, SD
= 1.38) had significantly lower masculinity scores
than did tops (M = 4.78, SD = 0.98) and versatiles (M
= 4.63, SD = 0.66) while tops and versatiles did not
differ from each other. Hypothesis 1 was therefore
supported.
H2: IPP and femininity. Another one-way
between-groups ANOVA was conducted to explore
the relation of IPP and femininity as measured by the
relevant subscale of the Bem (Table 2). There was a
statistically significant difference in scores between
the three groups [F (2, 67) = 6.75, p = .002]. Post-hoc
9
PSYCHOLOGICAL SCIENCE: Eric Carter
Table 2. Study variables by intercourse position
preference group.
Top (n = 30)
Versatile (n = 28)
Variable
M (SD) or N M (SD) or N (%)
(%)
Internalized
homophobia
3.17 (1.46)
Bem: Masculinity
4.78a (0.98)
a
b
2.46 (0.91)
Table 3. Bivariate correlations among key study
variables
Bottom (n = 12)
M (SD) or N (%)
b
2.37 (0.62)
F statistic
F (2, 67) = 3.54,
3.48b (1.38)
1 IH
1. Internalized
homophobia
1.00
2. Bem masculinity
.192
3. Bem femininity
F (2, 67) = 8.57,
p < .001
Bem: Femininity
4.26a (0.91)
4.67a (0.65)
3.53b (1.30)
4. BSI
F (2, 67) = 6.75,
p = .002
Brief Symptom
Inventory
0.47 (0.54)
0.38 (0.31)
0.67 (0.48)
1
5. Position preference
F (2, 67) = 1.68,
p = .194
2 Masc.
3 Fem.
4 BSI
5 IPP
1.00
(p = .112)
p = .034
4.63a (0.66)
Variable
-.236
.319
1.00
(p = .049)
(p = .007)
-.067
-.431
-.106
(p = .581)
(p < .001)
(p = .383)
-.257
-.440
-.270
.142
(p = .032)
(p < .001)
(p = .024)
(p = .240)
1.00
1.00
Note. Means sharing a letter in their superscript are not
significantly different at the .05 level on Fisher LSD posthoc tests.
Note. 1 Position preference was left in its continuous form;
with lower scores indicating a preference for top.
comparisons using the Fisher LSD test indicated that
the mean femininity scores for bottoms (M = 3.53, SD
= 1.30) was significantly lower than that of tops (M =
4.26, SD = 0.91) and versatiles (M = 4.67, SD = 3.53)
which did not differ significantly from each other.
Hypothesis 2 was therefore not supported.
H3: IPP and internalized homophobia. A
one-way between-groups ANOVA was conducted
to explore the relation between intercourse position
preference (IPP) and internalized homophobia (Table
2). There was a statistically significant difference in
mean scores for the three groups [F (2, 67) = 3.54, p
= .034]. Post-hoc comparisons using the Fisher LSD
test indicated that tops (M = 3.17, SD = 1.46) had
significantly higher internalized homophobia scores
than did versatiles (M = 2.46, SD = 0.91) and bottoms
(M = 2.37, SD = 0.62). Bottoms and versatiles did not
differ significantly from each other. Hypothesis 3 was
therefore supported.
H4: Femininity and global mental health.
To address the fourth hypothesis, Pearson productmoment correlations were computed between Bem
femininity scores and BSI global distress scores.
Results are summarized in Table 3. The correlation
between Bem femininity scores and BSI was r = -.106,
p = .383. Therefore, Hypothesis 4 was not supported.
On the other hand, Bem masculinity scores and the
BSI were significantly related, r = -.431, p < .001,
indicating higher endorsement of masculine gender
traits was associated with lower psychiatric distress.
H5: Internalized homophobia and global
mental health. A Pearson product-moment correlation
was computed between internalized homophobia, as
measured by the Reactions to Homosexuality Scale,
and psychiatric distress, as measured by the BSI global
score (Table 3). The resulting correlation indicated the
two were not significantly related, r = -.067, p = .581.
Furthermore, a one-way between-groups ANOVA
comparing the three IPP groups on mean BSI scores
was not significant (Table 2). Therefore, Hypothesis 5
was not supported.
Part 3: Post-hoc Analyses
Because of the significant relation found
between self-reported masculinity and internalized
homophobia, ANCOVAs were conducted to discover
if, after controlling for variance due to sex roles and
homophobia, IPP would explain some of the variability
in scores on the BSI. Table 4 provides a summary of
adjusted means by IPP group and specific analyses are
described next. A one way between-groups ANCOVA
was conducted to determine if psychiatric distress (as
measured by mean score on the BSI) was partially
explained by IPP, after controlling for internalized
homophobia. There was no significant difference
between BSI scores [F (2, 64) = 0.00, p = .999].
Another one way between-groups ANCOVA
was conducted to determine if psychiatric distress was
related to IPP, after controlling for Bem masculinity
scores. There was a significant difference between the
three groups [F (2, 64) = 3.87, p = .026]. The mean
10
PSYCHOLOGICAL SCIENCE: Eric Carter
Table 4. Study variables by intercourse position
preference group, with adjusted means.
Top (n =
30)
Versatile (n =
28)
Bottom (n =
12)
M (SD)
M (SD)
M (SD)
or N (%)
or N (%)
or N (%)
Brief Symptom
Inventory (adjusted
for internalized
homophobia)
0.48
0.38
0.64
F(2, 64) = 0.00,
p = .999
Brief Symptom
Inventory (adjusted
for Bem Masculinity)
0.56a
0.38b
0.53a
F (2, 64) = 3.87,
p = .026
Brief Symptom
Inventory (adjusted
for Bem Femininity)
0.48
0.42
0.59
F (2, 64) = 1.73,
p = .186
Variable
F statistic
Note. Means sharing a letter in their superscript are not
significantly different at the .05 level on Fisher LSD posthoc tests.
BSI score for versatiles (0.38) was significantly lower
than that of both tops (0.56) and bottoms (0.53). Tops
and bottoms did not differ from one another. Finally,
an ANCOVA examined if psychiatric distress was
related to IPP after controlling for Bem femininity
scores. The results were not significant [F (2, 64) =
1.73, p = .186].
Discussion
The study was designed to discover if anal
intercourse position preference (IPP) for men who
have sex with men (MSM) was related to sociodemographic characteristics, gender role orientation,
and internalized homophobia, and whether gender
role orientation and internalized homophobia were,
in turn, related to psychiatric distress. In the first set
of analyses, we noted few demographic differences
among IPP groups. Bottoms were the least likely
to be single, reported the fewest number of sexual
partners in the last year (average of one), and had the
lowest rates of condom use. It may be that, because
they were in stable long-term relationships, bottoms
reported lower condom use. Tops and versatiles were
very similar in terms of frequency of condom use
and number of past-year sexual partners (average of
two). Other than those differences, groupings based
on intercourse position preference had few statistical
associations with demographic characteristics; sexual
satisfaction, education level, annual income, alcohol
use, and pornography use were not statistically
significantly related to IPP. Taken together, it appears
that IPP is not associated with most
psychosocial variables. In fact, the only differences
that emerged suggest bottoms report more stable,
long-term relationships than tops and versatiles.
In our second set of analyses, we tested specific
hypotheses about how IPP related to gender roles and
psychological well-being. The first hypothesis (H1),
which predicted tops and versatiles would endorse
higher masculine traits than bottoms, received
full statistical support. Indeed, in MSM, being the
“penetrator” in a sexual encounter is associated with
other self-described masculine characteristics, such
as being a leader, an aggressor, and confident. This
suggests that one way in which masculinity may be
expressed by men is through their sexual behaviors.
However, our second hypothesis (H2), which
predicted that versatiles and bottoms would endorse
more feminine traits compared to tops, was rejected.
Instead, both tops and versatiles endorsed high levels
of femininity, while bottoms had lower levels of
femininity. Overall, it appears that bottoms reported
low endorsement of both masculine and feminine
traits—seeing themselves as neither powerful or
self-reliant nor compassionate or loyal. This may
suggest bottoms have lower self-esteem than tops
and versatiles, reporting fewer positive personality
traits of any gender type. It is also possible that the
Bem Sex Role Inventory’s items no longer adequately
capture the domains of masculinity and femininity.
As Hoffman & Borders (2001) suggest, it is possible
that many of the gender-specific traits of 40 years
ago are now seen as positive for everyone, regardless
of gender. For instance, it may be that people are
equally likely to see men and women as strong and
independent, loyal and sensitive to others’ needs.
Our next set of analyses examined whether
there were significant differences among the IPP groups
on internalized homophobia and psychiatric distress.
Hypothesis three (H3) predicted that internalized
homophobia would be lower in versatiles and bottoms
compared to tops; this hypothesis was fully supported.
Given that being the “penetrator” in sexual relations is
a behavior consistent with a masculine self-identity, it
is not surprising that men who were less comfortable
with their sexual orientation or felt less affiliation with
the gay community also endorsed greater interest in
and history of insertive anal sex. This is consistent
with other work in MSM showing men who posted
online advertisements seeking sexual encounters with
non-gay-identified men were more likely to be looking
11
PSYCHOLOGICAL SCIENCE: Eric Carter
for a “masculine” partner who would penetrate them
(i.e., a masculine top) (Downing & Schrimshaw,
2014). Presumably, these advertisements suggest that
bottoms who desire a top partner recognize the ideal
partner may have some internalized homophobia.
Finally, consistent with gender schema theory,
we expected that both femininity (H4) and internalized
homophobia (H5) would relate significantly to
psychiatric distress. However, neither of these
hypotheses was statistically supported. On the other
hand, we found masculinity was significantly related to
lower psychiatric distress. The findings of H4 and H5,
when viewed in the context of hegemonic masculinity
and patriarchy (see Connell, 2005), suggest that
masculinity may serve as a protective factor against
psychiatric distress in MSM. This is consistent with
a large body of literature demonstrating the benefits
of a more masculine gender role in both men and
women (Bassoff & Glass, 1982; Cella et al., 2013;
Cheng, 2005; Lefkowitz & Zeldow, 2006; Prakash et
al., 2010; Taylor & Hall, 1982; Woodhill & Samuels,
2003). In particular, while having sex with other men
may place a person’s masculinity in question, other
traits can compensate for this. For example, MSM
who are masculine in ways other than their sexual
behavior (such as being aggressive, strong leaders,
and analytical) can still reap many of the benefits of
being perceived as a member of the powerful in-group
of men. A more direct assessment of the potential
discrimination faced by MSM who are more feminine
and less masculine in their traits may help clarify the
relations between gender roles and psychiatric distress
we observed in the current study. In fact, one prior
study suggested discrimination experiences mediate
the relation between gender non-conformity and
psychiatric distress (Sandfort et al., 2007).
We had originally anticipated IPP would relate
to gender role orientation and internalized homophobia
directly, and indirectly to psychiatric distress. Once
we failed to find statistically significant associations
between gender roles or homophobia and psychiatric
distress, we explored whether IPP was directly related
to psychiatric distress. Of note, the only analysis that
found a relationship between these two variables was
a post-hoc ANCOVA that controlled for masculinity.
Findings suggest that versatiles had the best mental
health of all groups, with tops and bottoms reporting
similar (albeit low) psychiatric distress.
While the findings are preliminary, we think
they warrant further investigation. For instance,
perhaps versatile sexual behavior is comparable to an
androgynous gender role orientation in that it allows
for an expanded behavioral repertoire and flexibility
in partnering decisions, where decisions about
relationships can be made based on factors other than
sexual compatibility. It may also be that a versatile
identity is associated with other psychological
variables, such as openness to experiences, which
relate to positive mental health (Zoeterman & Wright,
2014).
Limitations and Future Directions
This study was novel in bringing together
variables that had only been studied in relative isolation:
anal intercourse position preference, internalized
homophobia, gender roles, and psychiatric distress.
The results largely conformed to expectations and
revealed a possible benefit to being versatile in IPP for
MSM, much as an androgynous gender role confers
benefits on individuals (Prakash et al., 2010; Woodhill
& Samuel, 2003).
However, the study’s strengths must be
considered within the context of its limitations.
Specifically, this study was limited by small sample
size (n = 70) and being an internet-only self-report
questionnaire study. In addition, the sample size was
limiting in terms of IPP diversity. Furthermore, the
compensation structure for participants changed as
the study progressed because of difficulties obtaining
participants. Although the number of bottoms in the
study (19%) was close to established percentages in
the other studies (21%) of MSM (Wei & Raymond,
2011), there were only 12 participants who were
categorized as bottoms, limiting statistical power and
sensitivity of analyses and generalizability of results.
In addition to addressing these methodological
concerns, future research should explore potential
variables that may explain the relations observed
in this study. For instance, studies examining how
discrimination experiences may mediate the relation
between low masculinity and high psychiatric distress
are warranted, as are studies that include assessment
of personality characteristics that may relate to
both sexual behavior (IPP) and mental health, such
as openness to experience. One variable not well
explored by this study was outness. Researchers (e.g.,
Scrimshaw et al., 2013) have found a relationship
between being “out” versus “in the closet” and mental
health, with outness associated with more positive
12
PSYCHOLOGICAL SCIENCE: Eric Carter
mental health. Furthermore, it may be that IPP changes
as a function of outness. The major life changes that
accompany the coming-out process (e.g., acceptance
versus stigmatization, a change in social group) would
be expected to have psychological impact. If IPP is a
proxy for outness, it could be outness and its sequelae
explain the relation between IPP and mental health.
Future studies could also explore IPP in cultures
outside of the United States and with larger samples
of diverse racial and ethnic groups within the United
States to determine if similar findings would occur in
more diverse populations of MSM.
References
Baggaley, R., White, R., & Boily, M. (2010).
HIV transmission risk through anal
intercourse: Systematic review, metaanalysis and implications for HIV
prevention. International Journal of
Epidemiology, 39, 1048-1063. doi:10.1093/
ije/dyq057
Bailey, J. M., Kim, P.Y., Hills, A., & Linsenmeier,
J. (1997). Butch, femme, or straight
acting? Partner preferences of gay men and
lesbians. Journal of Personality and Social
Psychology, 73, 960-973.
Bassoff, E. S., & Glass, G. V. (1982). The
relationship between sex roles and mental
health: A meta-analysis of twenty-six studies.
Counseling Psychologist, 10, 105-112.
Bem, S. L. (1974). The measurement of
psychological androgyny. Journal of
Consulting and Clinical Psychology, 42, 155162.
Bem, S. L. (1981). Gender schema theory:
A cognitive account of sex typing.
Psychological Review, 88, 354-364.
Cella, S., Iannaccone, M., & Cotrufo, P. (2013).
Influence of gender role orientation
(masculinity versus femininity) on
body satisfaction and eating attitudes in
homosexuals, heterosexuals and transsexuals.
Eating and Weight Disorders, 18, 115-124.
doi: 10.1007/s40519-013-0017-z
Cheng, C. (2005). Process underlying gender role
flexibility: Do androgynous individuals
know more or know how to cope? Journal of
Personality, 73, 645-673. doi: 10.1111/j.1467-
6494.2005.00324.x
Coleman, D., Kaplan, M. S., & Casey, J. T. (2011).
The social nature of male suicide: A new
analytic model. International Journal of
Men’s Health, 10, 240-252. doi: 10.3149/
jmh.1003.240
Connell, R. W. (2005). Masculinities (2nd ed.).
Berkeley, CA: University of California Press.
Derogatis, L. R., & Melisaratos, N. (1986). The Brief
Symptom Inventory: An introductory report.
Psychological Medicine, 13, 595-605.
Downing Jr., M. J., & Schrimshaw, E. W.
(2014). Self-presentation, desired partner
characteristics, and sexual behavior
preferences in online personal advertisements
of men seeking non-gay-identified men.
Psychology of Sexual Orientation and Gender
Diversity, 1, 30-39.
Gordon, A. R., & Meyer, I. H. (2007). Gender
noncomformity as a target of prejudice,
discrimination, and violence against
LGB individuals. Journal of LGBT
Health Research, 3, 55-71. doi:
10.1080.15574090802093562
Grov, C., Parsons, J. T., & Bimbi, D. S. (2010).
Sexual compulsivity and sexual risk in
gay and bisexual men. Archives of Sexual
Behavior, 39, 940-949. doi:10.1007/s10508009-9483-9
Hart, T. A., Wolitski, R. J., Purcell, D. W., Gómez,
C., & Halkitis, P. (2003). Sexual behavior
among HIV-positive men who have sex
with men: What’s in a label? Journal of Sex
Research, 40, 179-188.
Herek, G. M., Cogan, J. C., Gillis, J., & Glunt,
E. K. (1998). Correlates of internalized
homophobia in a community sample of
lesbians and gay men. Journal of the Gay &
Lesbian Medical Association, 2, 17-25.
Hoffman, R., & Borders, L. (2001). Twenty-five
years after the Bem Sex-Role Inventory:
A reassessment and new issues regarding
classification variability. Measurement &
Evaluation in Counseling & Development,
34, 39.
Jin, F., Crawford, J., Prestage, G. P., Zablotska, I.,
Imrie, J., Kippax, S. C., ... & Grulich, A.
E. (2009). Unprotected anal intercourse,
risk reduction behaviours, and subsequent
13
PSYCHOLOGICAL SCIENCE: Eric Carter
HIV infection in a cohort of homosexual
men. AIDS, 23, 243-252. doi:10.1097/
QAD.0b013e32831fb51a
Kippax, S., & Smith, G. (2001). Anal intercourse and
power in sex between men. Sexualities, 4,
413-436.
Koblin, B., Husnik, M., Colfax, G., Huang, Y.,
Madison, M., Mayer, K., ... & Buchbinder, S.
(2006). Risk factors for HIV infection among
men who have sex with men. AIDS, 20, 731739.
Lanzieri, N., & Hildebrandt, T. (2011). Using
hegemonic masculinity to explain gay male
attraction to muscular and athletic men.
Journal of Homosexuality, 58, 275-293. doi:1
0.1080/00918369.2011.540184
Lefkowitz, E. S., & Zeldow, P. B. (2006).
Masculinity and femininity predict optimal
mental health: A belated tandrogyny
hypothesis. Journal of Personality
Assessment, 87, 95-101. doi:10.1207/
s15327752jpa8701_08
Marrs, H., Sigler, E. A., & Brammer, R. D. (2012).
Gender, masculinity, femininity and help
seeking in college. Masculinities and
Social Change, 1, 67-292. doi: 10.4471/
MCS.2012.16
Mustanski, B. S., Newcomb, M. E., Du Bois,
S. N., Garcia, S. C., & Grov, C. (2011).
HIV in young men who have sex with
men: A review of epidemiology, risk and
protective factors, and interventions.
Journal of Sex Research, 48, 218-253. doi:
10.1080/00224499.2011.558645
Prakash, J., Kotwal, A., Ryali, V., Srivastava, K.,
Bhat, P., & Shashikumar, R. (2010). Does
androgyny have psychoprotective attributes?
A cross-sectional community-based study.
Industrial Psychiatry Journal, 19, 119-124.
doi:10.4103/0972-6748.90343
Rieger, G., & Savin-Williams, R. (2012). Gender
nonconformity, sexual orientation, and
psychological well-being. Archives of Sexual
Behavior, 41, 611-621. doi:10.1007/s10508011-9738-0
Ross, M., & Rosser, B. (1996). Measurement and
correlates of internalized homophobia: A
factor analytic study. Journal of Clinical
Psychology, 52, 15-21.
Safren, S. A., Reisner, S. L., Herrick, A., Mimiaga,
M. J., & Stall, R. D. (2010). Mental health
and HIV risk in men who have sex with men.
Journal of Acquired Immune Deficiency
Syndrome, 55, S74-77. doi: 10.1097/
QA1.0b013e3181fbc939
Sandfort, T. G. M., Melendez, R. M., & Diaz, R. M.
(2007). Gender nonconformity, homophobia,
and mental distress in Latino gay and bisexual
men. Journal of Sex Research, 44, 181-189.
doi: 10.1080/00224490701263819
Schrimshaw, E., Siegel, K., Downing, M., & Parsons,
J. (2013). Disclosure and concealment of
sexual orientation and the mental health of
non-gay-identified, behaviorally bisexual
men. Journal of Consulting and Clinical
Psychology, 81, 141-153. doi:10.1037/
a0031272
Smolenski, D. J., Diamond, P. M., Ross, M. W., &
Rosser, B. R. S. (2010). Revision, criterion
validity, and multi-group assessment of the
Reactions to Homosexuality Scale. Journal
of Personality Assessment, 92, 568-576. doi:
10.1080/00223891.2010.513300
Taylor, M. C., & Hall, J. A. (1982). Psychological
androgyny: Theories, methods, and
conclusions. Psychological Bulletin, 92, 347366. doi: 10.1037/0033-2909.92.2.347
Thomas, B., Mimiaga, M. J., Menon, S.,
Chandrasekaran, V. V., Murugesan, P. P.,
Swaminathan, S., ... & Safren, S. A. (2009).
Unseen and unheard: Predictors of sexual risk
behavior and HIV infection among men who
have sex with men in Chennai, India. AIDS
Education and Prevention, 21, 372-383.
doi:10.1521/aeap.2009.21.4.372
Van Druten, H., Van Griensven, F., & Hendriks,
J. (1992). Homoanal sex role separation:
Implications for analyzing and modeling the
spread of HIV. Journal of Sex Research, 29,
477-499.
Wegesin, D. J., & Meyer-Bahlburg, H. L. (2000).
Top/bottom self-label, anal sex practices,
HIV risk and gender role identity in gay men
in New York City. Journal of Psychology &
Human Sexuality, 12, 43-62. doi:10.1300/
J056v12n03_03
Wei, C., & Raymond, H. (2011). Preference for and
maintenance of anal sex roles among men
14
PSYCHOLOGICAL SCIENCE: Eric Carter
who have sex with men: Sociodemographic
and behavioral correlates. Archives of Sexual
Behavior, 40, 829-834. doi:10.1007/s10508010-9623-2
Weinrich, J. D., Grant, I., Jacobson, D. L., Robinson,
S., & McCutchan, J. (1992). Effects of
recalled childhood gender nonconformity on
adult genitoerotic role and AIDS exposure.
Archives of Sexual Behavior, 21, 559-585.
doi:10.1007/BF01542256
Williamson, I. (2000). Internalized homophobia
and health issues affecting lesbians and gay
men. Health Education Research, 15, 97-107.
Woodhill, B., & Samuels, C. A. (2003). Positive and
negative androgyny and their relationship
with psychological health and well-being. Sex
Roles, 48, 555-565.
Zoeterman, S. E., & Wright, A. J. (2014). The
role of openness to experience and sexual
identity formation in LGB individuals:
Implications for mental health. Journal
of Homosexuality, 61, 334-353. doi:
10.1080/00918369.2013.839919
15